The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands (start on Windows or open on macOS) to automatically launch the generated .drawio files after they are saved to the disk. Additionally, documentation provides command-line examples for exporting diagrams to PNG/SVG using the draw.io executable.
[EXTERNAL_DOWNLOADS]: The README files reference well-known and trusted external resources, including official websites and repositories for draw.io (diagrams.net) and Claude Code for installation and prerequisite purposes.
[PROMPT_INJECTION]: The skill processes untrusted user input to derive diagram content and filenames, creating a surface for indirect prompt injection.
Ingestion points: User-provided natural language descriptions of processes or systems processed in skill.md.
Boundary markers: None identified; user input is processed directly to generate the DiagramSpec and XML.
Capability inventory: File system write access for saving .drawio files and shell command execution for opening/exporting files.
Sanitization: The instructions attempt to sanitize output filenames by requiring the agent to follow a specific 'lowercase with hyphens' format.

anthropic-diagram